Dispersion penalty analysis for LED/single-mode fiber transmission systems

GREAT DEAL of attention has been focused recently on telecommunication systems based on single-mode fiber (SMF) and LED light sources for future deployment in the local network and subscriber loop. These systems combine the advantages of the low loss, large bandwidth, and upgrade potential of single-mode fiber with the high reliability and temperature stability, as well as the low cost of LED’s. Recent experiments have demonstrated the feasibility of LED-SMF ’ systems for transmission rates up to 560 Mbit/s and span lengths (at 140 Mbit/s) up to 50 km [1]-[lo]. Chromatic dispersion is a potential limitation at these bit rates and span lepgths, however, because of the broad spectral widths of LED’s. Even for a practical system of shorter transmission distance, the dispersion penalty remains an important consideration in the power budget since the total power coupled into single-mode fiber is relatively small for LED systems. It is thus important to have realistic estimates of dispersion penalties in LED-SMF system design. Penalties for chromatic dispersion in multimode fiber have been calculated for LED’s based on the assumptions of an optical receiver which has been re-equalized to compensate for the fiber dispersion, and an f receiver noise spectral density [ 111, [12]. In practical systems, however, these assumptions are often not applicable. In this paper we present a new calculation of dispersion penalties arising from intersymbol interference for LED-SMF systems with no re-equalization (the more commonly adopted scheme), and compare the results obtained with a treatment of the re-equalized system which considers both white, andf2 receiver noise spectral densities. Simple approximations of dispersion penalties are derived, and bitrate distance product limits are presented as a useful estimation tool for system feasibility studies. Single-mode fiber transmission experiments at 90, 140, and 560 Mbit/ s have been performed using 1.3- and 1.5-pm LED’s [3][6], and the measured dispersion penalties are compared with the results of the above analysis.